Science

A Highly Accurate Oscar Prediction Algorithm Announced Its Picks for 2017

IMDB

The 89th Academy Awards airs this Sunday, and if you want to impress at your Oscars party, listen up: Statistician Iain Pardoe, Ph.D. is the developer behind a computer algorithm that has accurately predicted 75 percent of Oscar wins since 2005, and he’s just released his latest forecast. Every Oscar season, Pardoe updates the model with new, relevant data and fires up the analysis again — meaning that this year, the model is likely to be more accurate than it’s ever been.

Pardoe’s model — which was 100 percent accurate in 2008, 2009, 2013, and 2014 — focuses on four categories: Best Picture, Best Director, Best Actor, and Best Actress. His computer model frames the question of Oscar prediction as a “discrete choice problem,” an economic model based on the idea that there are specific patterns in choices. Over the years, Pardoe has found that certain external factors can predict a win. For example, an actor with numerous previous wins has a lower chance of winning again, while movies gunning for Best Picture are more likely to win if they are nominated in other categories. Pardoe doesn’t run his analysis until he can input important data from the Producers Guild of America Awards — but he also believes Golden Globe wins are also a strong predictor of what will go down at the Oscars.

While upsets can always happen, Pardoe’s model seems to work: When the model was applied to nominations from 1938 to 2015, it correctly predicted 222 out of 312 awards. It most often gets Best Director right (83 percent accuracy) and Best Actress wrong (58 percent accuracy). According to this year’s analysis, these are the films, directors, and stars likely to win big in 2017:

Best Picture: La La Land

Two attractive people in love.

Giphy

This model has the musical romance La La Land winning with a 97 percent probability. The coming-of-age drama Moonlight comes in second, with a 2 percent probability.

Best Director: Damien Chazelle

Director Damien Chazelle of 'La La Land' speaks onstage.

Getty Images / Matt Winkelmeyer

The La La Land director leads the pack according to the model, set to win with a 98 percent probability. Possible Suicide Squad 2 director Mel Gibson and Moonlight’s Barry Jenkins are tied for second most likely winner, each with a 1 percent probability of winning.

Best Actor in a Leading Role: Ryan Gosling

Play those keys, Ryan.

Giphy

Pardoe’s model has Gosling, who stars as Sebastian in La La Land in the lead — but not by much. Gosling has a 44 percent chance of winning, while Casey Affleck of Manchester by the Sea clocks in at 29 percent and Denzel Washington of Fences at 4 percent.

Best Actress in a Leading Role: Emma Stone

Emma Stone as Mia. 

Giphy

Emma Stone, who plays the role of Mia in La La Land, is by far the most favored to win this category: The computer model predicts she has a 94 percent chance of winning. Other possible victors are Isabelle Huppert, who stars in Elle (4 percent), and Ruth Negga, who plays a leading role in Loving (1 percent). Previous winners Natalie Portman and Meryl Streep, writes Pardoe, are “not really in the runnings to win again.”

Related Tags